multiple modes of binding enhance the affinity of dc-sign for high
TRANSCRIPT
Feinberg et al.
1
MULTIPLE MODES OF BINDING ENHANCE THE AFFINITY OF DC-SIGNFOR HIGH-MANNOSE N-LINKED GLYCANS FOUND ON VIRAL
GLYCOPROTEINSHadar Feinberg1, Riccardo Castelli2, Kurt Drickamer3, Peter H. Seeberger2, and William I.
Weis1
From the 1Departments of Structural Biology and Molecular & Cellular Physiology, StanfordUniversity School of Medicine, Stanford, CA 94305 USA, 2Laboratory for Organic Chemistry,
Swiss Federal Institute of Technology (ETH) Zurich, Wolfgang Pauli Strasse 10, CH 8093Zurich, Switzerland, and 3Division of Molecular Biosciences, Biochemistry Building, Imperial
College, London SW7 2AZ, United KingdomRunning title: DC-SIGN binding to high-mannose oligosaccharides
Address correspondence to William I. Weis, Dept. of Structural Biology, Stanford UniversitySchool of Medicine, 299 Campus Drive West, Stanford, CA 94305 USA Tel +1 650 725 4623;Fax +1 650 723 8464; email: [email protected]
The dendritic cell surface receptor DC-SIGNand the closely related endothelial cell receptorDC-SIGNR specifically recognize high-mannose N-linked carbohydrates on viralpathogens. Previous studies have shown thatthese receptors bind the outer trimannosebranch Manα1-3[Manα1-6]Manα- present inhigh-mannose structures. Although thetrimannoside binds to DC-SIGN or DC-SIGNRmore strongly than mannose, additional affinityenhancements are observed in the presence ofone or more Manα1-2Manα- moieties on thenon-reducing termini of oligomannosestructures. The molecular basis of thisenhancement has been investigated bydetermining crystal structures of DC-SIGNbound to a synthetic six-mannose fragment of ahigh-mannose N-linked oligosaccharide,Manα1-2Manα1-3[Manα1-2Manα1-6]Manα1-6Man, and to the disaccharide Manα1-2Man.The structures reveal mixtures of two bindingmodes in each case. Each mode features typicalC-type lectin binding at the principal Ca2+
binding site by one mannose residue. Inaddition, other sugar residues form contactsunique to each binding mode. These resultssuggest that the affinity enhancement displayedtowards oligosaccharides decorated with theManα1-2Manα- structure is due in part tomultiple binding modes at the primary Ca2+
site, which provide both additional contacts anda statistical (entropic) enhancement of binding.
The dendritic cell receptor DC-SIGNfunctions in the initial recognition of pathogens,
and also in adhesive interactions with T cells thatscan the surface of dendritic cells forcomplementary peptide antigen–MHC complexes(1,2). Although the epitope for T cell interactionshas not been defined, interactions with pathogensexploit the ability of DC-SIGN to recognize bothbranched fucosylated structures bearing terminalgalactose residues, and high-mannose N-linkedoligosaccharides (3-5). The latter specificityallows DC-SIGN to act as a receptor for severalenveloped viruses that bear high-mannosestructures on their surface glycoproteins, mostnotably human immunodeficiency virus (HIV4) (6-8). A related receptor found on endothelia in theliver, lymph nodes, and placenta, designated DC-SIGNR or L-SIGN, does not recognizefucosylated carbohydrates but shares with DC-SIGN the ability to bind tightly to high-mannoseN-linked carbohydrates and to serve as a viralreceptor (5,6,9-11).
DC-SIGN and DC-SIGNR are membersof the C-type lectin family of Ca2+-dependentcarbohydrate-binding proteins. The two receptorshave similar primary structures, each of whichcomprises a short N-terminal cytoplasmic tail, atransmembrane anchor, a tetramerization domain,and a C-terminal carbohydrate-recognition domain(CRD) (6). Crystal structures of the DC-SIGNand DC-SIGNR CRDs reveal the typical long-form C-type lectin fold (4). There are three Ca2+
seen in these structures, of which one, designatedthe principal Ca2+, is common to all C-type lectins.The hallmark of sugar binding to C-type lectins isthe direct coordination of the principal Ca2+ byvicinal hydroxyl groups of a pyranose ring, which
http://www.jbc.org/cgi/doi/10.1074/jbc.M609689200The latest version is at JBC Papers in Press. Published on December 6, 2006 as Manuscript M609689200
Copyright 2006 by The American Society for Biochemistry and Molecular Biology, Inc.
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
2
also form hydrogen bonds with the amino acidside chains that serve as the other Ca2+ ligands(12). In the case of mannose-like ligands, vicinal,equatorial 3- and 4-OH groups form thesecoordination and hydrogen bonds. Specificity forparticular oligosaccharides comes from additionalcontacts made to flanking regions of the C-typeCRD.
Competition assays in which a test ligandis used to compete radiolabelled mannose-BSAfrom immobilized CRD have been used toexamine the relative affinities of mannose-containing structures for DC-SIGN and DC-SIGNR (4,6). The trimannose core structureManα1-3[Manα1-6]Man was found to bind 4-foldbetter than mannose to DC-SIGN and 2-fold betterto DC-SIGNR, and the disaccharide Manα1-2Manshows similar preferences (4). A pentasaccharidecorresponding to the inner five mannoses of ahigh-mannose oligosaccharide but lacking allterminal α1-2 linked mannoses binds 7- and 4-foldbetter than mannose to DC-SIGN and DC-SIGNR.The full N-linked high mannose oligosaccharideMan9GlcNAc2, however, shows much moresubstantial affinity enhancements (4). These datasuggested that the presence of the Manα1-2Manmoieties at the non-reducing termini of high-mannose oligosaccharides might providesubstantial affinity enhancements, perhaps byinteracting with a secondary binding site for thisgroup. The surface glycoproteins of HIV andother enveloped viruses are relatively rich in Man8
and Man9 structures (13), so high affinity bindingto such glycans contributes to selective interactionof DC-SIGN and DC-SIGNR with thesepathogens.
Here, the mechanism by which terminalManα1-2Man groups enhances affinity towardsDC-SIGN and DC-SIGNR is investigated usingsynthetic fragments of the full N-linked highmannose structure in binding and structuralstudies. The data indicate that multiple modes ofbinding at the DC-SIGN carbohydrate-binding siteprovide a statistical enhancement of the affinitybut do not account for all of the observed affinitydifferences. The different binding orientationsfeature contacts between the terminal mannose anddifferent regions of the proteins, which likely
provide the remaining component of the increasedaffinity for larger glycans.
Experimental Procedures
Protein expression – The DC-SIGN carbohydrate-recognition domain was expressed in Escherichiacoli as described (6) and used for cocrystallizationwith Manα1-2Man. A similar construct lackingthe C-terminal 12 residue extension was used forcocrystallization with Man6. Both proteins werepurified as described (6).
Synthesis and purification of Man6 and Man9
oligosaccharides – Compounds Man9 and Man6a
were prepared analogously to those previouslydescribed in the literature (14,15). O - M eprotection at the reducing end was chosen todiminish the possible interference of the linkerwith the binding site of the protein. CompoundMan6b was prepared following the same approachas Man9, using Methyl 2,3,4-Tri-O-benzyl-α-D-mannopyranoside (16,17) as the core sugar unit.After removal of all protecting groups, thecompounds were dialyzed two times each for 12hours against 2 L of Millipore water, thenlyophilized. The Manα1-2Man disaccharide waspurchased from Sigma.
Binding assays – Solid phase competition bindingassays were performed using bacterially expressedCRDs of DC-SIGN and DC-SIGNR, with 125I-Man-bovine serum albumin employed as thereporter ligand (6). Assays were performed atleast twice in duplicate, except that theMan9GlcNAc2 glycan was assayed only once induplicate because only limited quantities wereavailable. Sugar concentrations were determinedusing the anthrone reaction (18).
Crystallization and data collection – Crystals ofDC-SIGN CRD complexed with Man2 or Man6b
(Fig. 1) were grown at 21ºC by hanging dropvapor diffusion (1 µL protein to 1 µL reservoir ina drop for Man2 and 0.6:0.6 for Man6b). Theprotein solution that gave crystals for the DC-SIGN/Man2 complex contained 10 mg ml-1
protein, 5 mM CaCl2 and 25 mM Man2. Theprotein solution that gave crystals for the DC-SIGN/Man6 complex contained 5 mg ml-1 protein,5 mM CaCl2 and 50 mM Man6b. The reservoir
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
3
solution for both crystals contained 30% (w/v)polyethylene glycol 3000, 0.2 M NaCl, 0.1 M TrispH 7.0. Crystals were transferred to a drop ofreservoir solution with added sugar, were frozen inliquid nitrogen and maintained at 100 K duringdata collection. DC-SIGN–Man6b complex datawere measured on an ADSC Q315 CCD detectorat beam line 11-1 of the Stanford SynchrotronRadiation Laboratory. DC-SIGN–Man2 diffractiondata were measured on an ADSC Q315 CCDdetector at beam line 5.0.2 of the Advanced LightSource. Diffraction data were processed withMOSFLM and SCALA (19) (Table 1).
Structure determination – Crystals of both theMan6b and Man2 complexes were essentiallyisomorphous to the previously determined DC-SIGN-CRD–Man4 complex (5), even though thelatter was obtained using slightly differentcrystallization conditions. The asymmetric unitcontains one copy of the protein-ligand complex.Rather than reindexing to allow direct rigid bodyrefinement, the two structures were determined bymolecular replacement with the DC-SIGN CRDmodel from the Man4 complex (Protein Data BankID 1SL4). The Man2 complex structure wasdetermined with the program MOLREP (19),which gave a correlation coefficient of 70% and Rvalue of 33% for data to 3Å. The Man6b complexwas solved with program COMO (20), which gavea correlation coefficient of 56% and R of value31% for data to 3.5Å. Refinement and mapcalculations for both structures were performedwith CNS (21). The maximum-likelihoodamplitude target was used, with bulk solvent andanisotropic temperature factor corrections appliedthroughout the refinement. As refinementprogressed it became clear that the ligand is boundto the site in two alternative, overlappingorientations in both structures. The twoconformations were assigned occupancies of 75%and 25% based on the quality of the electrondensity and refined temperature factors. For eachligand orientation, Figure 2 shows the Fo-Fcelectron density calculated from coordinatesomitting the indicated orientation but including theother. Water molecules were added to peaks>3σ in Fo-Fc maps that were within hydrogen bonddistance to protein, sugar or other watermolecules. The final model of the DC-SIGNCRD–Man2 complex contains residues 253-384 of
DC-SIGN, two alternative conformations of thecarbohydrate ligand, 3 Ca2+, and 135 watermolecules. The final model of the DC-SIGNCRD–Man6 complex contains residues 253-384 ofDC-SIGN, two alternative conformations of theligand, 3 Ca2+, and 59 water molecules.Refinement statistics are presented in Table 1.
Results
Relative affinities of oligomannose structures forDC-SIGN and DC-SIGNR – To examine thecontribution of the Manα1-2Man groups presenton the termini of high-mannose oligosaccharide toDC-SIGN and DC-SIGNR binding, three syntheticoligomannose structures corresponding tofragments of Man9GlcNAc2 were tested in thecompetition assay. Man9 is the full 9-mannosestructure that would be linked to GlcNAc-GlcNAc-Asn in high-mannose N-linkedcarbohydrates (Fig. 1, green box). Man6a lacks thethree terminal mannoses of Man9 (Fig. 1, red box).Man6b is the substructure of Man9 that lacks theα1-3 branch arm (Fig. 1, blue box). Thesecompounds were assayed relative to mannose, andthe full Man9GlcNAc2 structure purified fromsoybean agglutinin was also included for directcomparison. The two Man6 structures bindsimilarly, with a 9- to 14-fold affinityenhancement relative to mannose, whereas theMan9 compound binds roughly twice as stronglyas the Man6 glycans. The full Man9GlcNAc2
glycan consistently shows 2 to 3-fold strongerbinding than Man9 (Table 2).
Structure of Man6b bound to DC-SIGN –Crystallization trials of complexes between Man9,Man6a and Man6b with the DC-SIGN CRD yieldedco-crystals with the 50 mM Man6b. The structureof this complex was determined at 2.4 Åresolution. The protein structure is identical to thatpreviously described for complexes withMan3GlcNAc2 (GlcNAcβ1-2Manα1-3[GlcNAcβ1-2Manα1-6]Man) (4) and Man4 (Manα1-3[Manα1-6]Manα1-6Man) (5). The ligand is bound in twooverlapping orientations, in a mixture estimated at75%, designated the major orientation (Fig. 3a,b)and 25%, designated the minor orientation (Fig.3c,d). Of the six mannoses in the compound, onlythree are visible in the major orientation (Manα1-
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
4
2Manα1-3Man) and two in the minor orientation(Manα1-2Man).
The penultimate α1-3-linked mannose thatforms one arm of the outer branched trimannoseunit (i.e., Manα1-2Manα1-3[Manα1-2Manα1-6]Manα1-6Man) binds to the primary Ca2+ site,and was similarly observed to bind to the Ca2+ inboth the DC-SIGN/Man4 (5) and DC-SIGN/Man3GlcNAc2 (4) structures. The majororientation corresponds to the arrangement seen inthese earlier crystal structures (Fig. 3e). The α1-6branch of the oligosaccharide is not visible,however, which is surprising given the shapecomplementarity and specific hydrogen bondsbetween the α1-6-linked mannose and Phe313,Ser360, and other residues in DC-SIGN that wereobserved in the earlier structures (4,5). The reasonfor this difference is obscure, especiallyconsidering the fact that the crystal is essentiallyisomorphous to the Man4 complex (5). In theMan4 and Man3GlcNAc2 structures, the α1-6-linked mannose has higher temperature factorsthan the α1-3-linked mannose, suggesting that itmay be more weakly bound. The α1-2-linkedmannose at the non-reducing terminus directlycontacts Val351.
In the second, less populated orientation,the same mannose residue is bound to the Ca2+, butits orientation is reversed by a 180° rotation abouta line bisecting the pyranose ring through the C3-C4 bond. This rotation exchanges the position ofthe 3- and 4-OH groups so that they still form theCa2+ coordination and hydrogen bondscharacteristic of C-type lectin-mannoseinteractions (Fig. 3b,d). A similar situation wasobserved in complexes of mannose-bindingproteins with various ligands (22). In thisorientation, only two sugars are visible: themannose at the Ca2+ site, and the non-reducingterminal α1-2-linked mannose, which formshydrogen bonds with Glu358, Ser360, and which alsointeracts with the face of the Phe313 ring (Fig. 3c).Thus, it appears that the Phe313 side chain hasimportant roles in the recognition of ligand ineither orientation. As only Manα1-2Man isvisible, it is not possible to distinguish whetherthese residues correspond to the α1-3 or α1-6arms of Man6b, or if they represent a mixture of
both (Fig. 1). An overlay of the major and minororientations is shown in Fig. 3f.
Structure of Manα1-2Man bound to DC-SIGN – Inorder to assess if DC-SIGN might have additionalsubsites for the Manα1-2Man residues found atthe non-reducing termini of high-mannoseoligosaccharides, the CRD was co-crystallizedwith 25 mM Manα1-2Man. The structure of thedisaccharide complex reveals binding only in theprincipal Ca2+ site; no other carbohydratemolecules were observed even at low electrondensity map contour levels. Manα1-2Man binds atthe principal Ca2+ site in two orientations, againrelated by a 180° rotation about the C3-C4 bondbisector. The major orientation is virtuallyidentical to that of the Manα1-2Man moiety in theminor Man6 ligand orientation and forms the samecontacts with DC-SIGN, including the contactwith Phe313 (Fig. 4a,b). In the minor orientation,only a single sugar is visible and forms the typicalCa2+ coordination and hydrogen bonds (Fig. 4c,d).This mannose is oriented identically to the Ca2+-bound mannose in the major Man6 orientation. Theelectron density maps, however, do not make clearif the sugar bound at the Ca2+ is the reducing ornon-reducing end of the disaccharide; it is possiblethat there is a mixture of the two. In particular,unlike the Man6 structure, the non-reducing α1-2-linked mannose is not visible. The electron densityfor this sugar is not especially well defined in theMan6b complex, so the lack of density for thisresidue in the Man2 complex could be due to itslow occupancy. It is also possible that thefavorable interaction with Val351 seen in the Man6
structure does not compensate for the loss ofentropy required to form this contact in thedisaccharide.
Models of Man9 binding – In order to assesswhether the two modes of binding observed in theMan6b and Man2 structures are relevant to a full, 9-mannose oligosaccharide, coordinates forMan9GlcNAc2 obtained by NMR analysis of thefree glycan (23) were superimposed on the twoorientations of Man6b. As noted previously (4),superposition of the outer branched trimannosemoiety reveals no significant steric clashesbetween the rest of the oligosaccharide and theprotein (Fig. 5a,b), with only the side chainrotamers of Leu371 needing adjustment to avoid
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
5
clashes. Superposition of the second orientation,which places the terminal α1-2-linked mannose ofthe outer α1-3 arm near Phe313, also reveals noclashes with the protein, with the possibleexception of interference between Arg345 and theinnermost GlcNAc residue (Fig. 5c). Furthermodeling, in which the terminal Manα1-2Mangroups of the other arms were superimposed,shows no steric clashes (Fig. 5d,e). Similar resultswere obtained with crystallographic coordinates ofMan9GlcNAc2 derived from the complex with aneutralizing anti-HIV antibody (24), although inthis case some minor adjustments to thecarbohydrate were required when the outerbranched mannose units were superimposed (datanot shown).
Discussion
The structure of the Man6b-DC-SIGN complexreveals two significantly populated binding modesfor the ligand. The major binding modecorresponds to that observed in previous crystalstructures, featuring a specific site for the outerbranched trimannose unit of high-mannose N-linked carbohydrates. In this orientation,additional contacts are formed between a non-reducing α1-2-linked terminal mannose andVal351. This region of DC-SIGN is also importantin binding to fucosylated sugars, and a terminalGlcNAc in the GlcNAc2Man3 complex (4). Thelatter compound binds 17-fold more strongly toDC-SIGN relative to mannose and 4-fold morethan the trimannose core, suggesting that theadditional interactions contribute significantly tooverall specificity. Surprisingly, a second bindingorientation was observed in which the mannose atthe principal Ca2+ site is reversed, therebygenerating new interactions between the non-reducing terminal mannose and the region aroundPhe313. Modeling indicates that this orientationwould be able to bind to the protein as part of afull 9-mannose structure (Fig. 5c-e).
The relevance of the dual binding modesof the Man6b compound was confirmed by thestructure of the Manα1-2Man disaccharide, whichshows the same interactions. In this case, thepreferred binding mode leaves the non-reducingend near the Phe313 site. This probably reflects a
different energetic balance of the Man2 and Man6
compounds, but in any case it is clear that thisbinding mode can be significantly populated. Thefact that no other binding sites for this ligand wereobserved suggests that there are no othersecondary subsites that interact with terminalManα1-2Man disaccharides in larger N-linkedhigh mannose oligosaccharides that might accountfor enhanced binding to such glycans.
The observation of dual binding modes,each resulting in unique contacts with DC-SIGN,permits a semi-quantitative explanation of theaffinity enhancements observed when high-mannose structures are decorated with α1-2-linkedmannose resides at the non-reducing termini.Using [P] and [L] to denote the concentrations offree protein and ligand, respectively, and [PLn] forthe concentration of the nth distinct protein–ligandcomplex with an association constant Kan, themeasured affinity constant is related to the affinityconstants of the individual binding modes by theequation:
Kameas = ([PL1] + [PL2] +…) / [P][L]
= Ka1 + Ka2 + …
Using the definition xn = Kan/Ka1 = [PLn] / [PL1],this relationship can be restated as:
Kameas = Ka1(1 + x2 + …)
Because ΔGmeas = -RTlnKameas
= -RTlnKa1 – RTln(1+ x2 + … )
= ΔG1 – RTln(1+ x2 + …),
ΔG1 = -RTlnKa1, and ΔGn = -RTlnKan ,
it follows that
xn = exp ( (-ΔGn - ΔG1) / RT).
Thus, for two binding modes of equal energy, ΔG2
- ΔG1 = 0, x2 = 1, and ΔGmeas = ΔG1 – RTln2, sothe ability to bind in two equally energetic modesprovides an additional RTln2 of free energy,corresponding to 0.41 kcal mol-1 at 25°C. For nisoenergetic binding modes the observedassociation constant will be n times the intrinsicassociation constant, whereas additional weakerbinding modes will increase the association by lessthan a factor of n. The effect of this statisticalfactor can be illustrated by comparing the binding
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
6
of Manα1-2Man with the biding of mannose.With free 3- and 4-OH groups, mannose couldbind in either of two orientations related by a 180°rotation that interchanges the 3- and 4-OH groups,as described above. In the disaccharide, eachresidue can in principle bind in either orientation,giving a total of four binding modes. Thus, if allof these modes were strictly equivalent, therelative Ka for the disaccharide would be twicethat of mannose. This argument ignores thepossible contribution of an alternative bindingarrangement involving the 1- and 2-OH groups,which has only been observed in the case ofgalactose binding to mannose-binding C-typelectins (25). The predicted ratio is seen for DC-SIGNR, but the ratio is about 4 for DC-SIGN,which probably indicates that there are additional,favorable interactions with DC-SIGN made by thesecond sugar of the disaccharide. It is alsopossible that free mannose can bind in only one oftwo modes, as seen in crystal structures ofmannose-binding prote ins bound tomonosaccharides, which generally show a singleorientation rather than a mixture in the binding site(22,25).
For more complex ligands, we canconsider the trisaccharide-binding mode observedin Man3GlcNAc2 and Man4 as a relatively highaffinity mode. The five-mannose core of the fullMan9 structure, which lacks all Manα1-2Mangroups, binds to DC-SIGN 7-fold better thanmannose and 4-fold better in the case of DC-SIGNR. Man5 also possesses the inner branchedtrimannose unit in the core which, in the absenceof the β-linked GlcNAc (4), might also bind. Theability to bind to either the inner or outer branchedtrimannose units likely explains the enhancementof Man5 over Man3 (4). Man6b binds 14-fold betterthan mannose to DC-SIGN and 12-fold better thanmannose to DC-SIGNR (Table 2). Man6b lacks theinner trimannose, but the outer branchedtrimannose binding mode and the “reversed” modein which the non-reducing terminal Man is boundnear Phe313, are present. The 2- and 4-foldenhancement of Man6b over Man5 binding to DC-SIGN and DC-SIGNR, can be accounted for inpart by the second binding mode. In the majororientation, the α1-2-linked terminal mannose onMan6b forms additional interactions relative toMan5, which might make this orientation
inherently stronger. In principle the Manα1-2Mangroup present on the termini of both the α1-3 andα1-6 branches of Man6b can bind in this secondorientation, and but they cannot be distinguishedin the structure (see Results), potentially providingthree distinct modes at the principal Ca2+ site(when the outer branched trimannose moiety plusthe two termini of the two branches areconsidered).
Given the unequal occupancies of the twoorientations observed in the Man6b complex, it islikely that an inherently stronger interaction of themajor, trimannose-binding mode, and thestatistical effect of the second mode, bothcontribute to the observed affinity enhancements.If the major orientation is of higher affinity thanthe non-reducing Manα1-2Man binding mode, theaffinity enhancements provided by the latter willbe less than a factor of n modes. If we assume thatthe observed occupancies reflect the relativeaffinities of the two binding modes, with theestimated 3:1 ratio of occupancy in both the Man6
and Man2 structures, x = 1/3, so ΔΔG = 0.65 kcalmol-1 would be the energy difference betweenthese modes. If we further assume that theobserved minor mode an equal mixture of the twodifferent α1-2 termini, then the ratios are3:0.5:0.5, and the energy difference is 1.06 kcalmol-1. This result illustrates that small differencesin energy due to differences in contacts combinedwith entropy losses due to conformationalimmobilization can give rise to preferred bindingorientations and likely explains why not allpossible modes are allowed even thought thebinding site requires only vicinal, equatorial OHgroups for Ca2+ ligation.
The affinities of Man6a for DC-SIGN andDC-SIGNR are enhanced to a similar extent as forMan6b. In this case, the outer branched trimannoseunit is present, but no Manα1-2Man moieties areappended to these branches. However, twoManα1-2Man groups present on the α1-3 arm ofthe inner branched trimannose would provide twomore binding modes. It is also possible that theinner branched trimannose unit could bind. Thus,this compound would appear to have a similarnumber and kind of binding modes as Man6b,despite their different covalent structures. In thefull Man9 structure, the inner and outer branched
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
7
trimannose units are present, as well as theManα1-2Man groups attached to the branches. Ifwe assume that the Manα1-2Manα1-2Man on theα1-3 branch of the inner trimannose structure canprovide two more modes, we have a total of sixmodes, which would explain its furtherenhancement relative to Man6.
Although this analysis makes severalassumptions about which modes of binding mightor might not occur, it is clear that the ability ofhigh-mannose oligosaccharides to interact withDC-SIGN and DC-SIGNR in multiple orientationscan give rise to statistical affinity enhancementsthat are consistent with the measured values.Energetic differences amongst the differentbinding modes also play an important role indetermining the affinity of each compound.Nonetheless, the 2 to 3-fold increase in affinitydisplayed by the full Man9GlcNAc2 structureversus Man9 is difficult to understand. Perhaps theinner GlcNAc residues restrict the conformation ofnearby sugar groups such that there is a smallerentropy penalty for binding, or alternatively, novelcontacts are formed between these residues andthe surface of the protein.
DC-SIGN and DC-SIGNR serve asreceptors for HIV and several other envelopedviruses by binding to the high-mannoseoligosaccharides present on viral surfaceglycoproteins. The CRD of DC-SIGN specificallyrecognizes an internal portion of the carbohydrate,namely the outer branched trimannnose unitunique to these carbohydrates. The presence ofManα1-2Man enhances the affinity of
oligomannose towards these receptors, eventhough by itself this disaccharide binds onlyslightly more strongly than mannose. The CRDhas an intrinsically high affinity for oligomannosestructures, and tetramerization likely providesfurther avidity enhancements for arrays of suchstructures (6,26). The ability of DC-SIGN andDC-SIGNR to bind high mannose glycans inmultiple orientations may facilitate thismultivalent binding of clusters of CRDs to glycansdisplayed in various arrangements on the surfaceof the virus, as proposed previously for cellsurface recognition by mannose-binding proteins(22). There are some parallels with the mechanismby which a neutralizing antibody to HIV, 2G12,binds specifically to the terminal Manα1-2Mangroups present on high-mannose carbohydrates(13,27). The binding site of 2G12 appears torecognize specifically a single orientation ofManα1-2Man present on the non-reducing terminiof Man9, but at least two of the three branchtermini bind to this antibody, which wouldcontribute some statistical enhancement ofaffinity. High avidity is provided in this case bythe unusual domain-swapped dimeric antibodystructure, which is proposed to displayappropriately spaced binding sites that match thespacing of these structures on the viral surface(24,28).
Acknowledgements – We thank Sofiya Fridman fortechnical assistance, and Dawn Torgersen andBrian Matthews for preparation of the glycansfrom soybean agglutinin.
References
1. Cambi, A. C., and Figdor, C. G. (2003) Curr. Opin. Cell Biol. 15, 539-5462. Steinman, R. M. (2000) Cell 100, 491-4943. Appelmelk, B. J., van Die, I., van Vliet, S. J., Vandenbroucke-Grauls, C. M. J. E., Geijtenbeek, T.
B. H., and van Kooyk, Y. (2003) J. Immunol. 170, 1635-16394. Feinberg, H., Mitchell, D. A., Drickamer, K., and Weis, W. I. (2001) Science 294, 2163-21665. Guo, Y., Feinberg, H., Conroy, E., Mitchell, D. A., Alvarez, R., Taylor, M. E., Weis, W. I., and
Drickamer, K. (2004) Nat. Struct. Molec. Biol. 11, 591-5986. Mitchell, D. A., Fadden, A. J., and Drickamer, K. (2001) J. Biol. Chem. 276, 28939-289457. van Kooyk, Y., and Geijtenbeek, T. B. H. (2003) Nat. Rev. Immunol. 3, 697-709
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
8
8. van Kooyk, Y., Appelmelk, B., and Geijtenbeek, T. B. H. (2003) Trends Mol. Med. 9, 153-1599. Alvarez, C. P., Lasala, F., Carrillo, J., Muñiz, O., Corbí, A. L., and Delgado, R. (2002) J. Virol.
76, 6841-684410. Pöhlmann, S., Baribaud, F., and Doms, R. W. (2001) Trends Immunol. 22, 643-64611. Simmons, G., Reeves, J. D., Grogan, C. C., Vandenberghe, L. H., Baribaud, F., Whitbeck, J. C.,
Burke, E., Burchmeier, M. J., Soilleux, E., Riley, J. L., Doms, R. W., Bates, P., and Pohlmann, S.(2003) Virology 305, 115-123
12. Drickamer, K. (1999) Curr. Opin. Struct. Biol. 9, 585-59013. Scanlan, C. N., Pantophlet, R., Wormald, M. R., Saphire, E. O., Stanfield, R., Wilson, I. A.,
Katinger, H., Dwek, R. A., Rudd, P. M., and Burton, D. R. (2002) J. Virol. 76, 7306-732114. Ratner, D. M., Adams, E. W., Su, J., O'Keefe, B. R., Mrksich, M., and Seeberger, P. H. (2004)
Chembiochem 5, 379-38215. Ratner, D. M., Plante, O. J., and Seeberger, P. H. (2002) Eur. J. Org. Chem. 2002, 826-83316. Ding, X., Wang, W., and Kong, F. (1997) Carbohydr. Res. 303, 445-44817. Sondheimer, S. J., Eby, R., and Schuerch, C. (1978) Carbohydr. Res. 60, 187-19218. Spiro, R. G. (1966) Methods Enzymol 8, 3-2619. Collaborative Computational Project, N. (1994) Acta Cryst. D50, 760-76320. Tong, L. (1996) Acta Cryst. A52, 782-78421. Brünger, A. T., Adams, P. D., Clore, G. M., Gros, P., Grosse-Kunstleve, R. W., Jiang, J.-S.,
Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G.L. (1998) Acta Cryst. D54, 905-921
22. Ng, K. K.-S., Kolatkar, A. R., Park-Snyder, S., Feinberg, H., Clark, D. A., Drickamer, K., andWeis, W. I. (2002) J. Biol. Chem. 277, 16088-16095
23. Woods, R. J., Pathiaseril, A., Wormald, M. R., Edge, C. J., and Dwek, R. A. (1998) Eur. J.Biochem. 258, 372-386
24. Calarese, D. A., Scanlan, C. N., Zwick, M. B., Deechongkit, S., Mimura, Y., Kunert, R., Zhu, P.,Wormald, M. R., Stanfield, R. L., Roux, K. H., Kelly, J. W., Rudd, P. M., Dwek, R. A., Katinger,H., Burton, D. R., and Wilson, I. A. (2003) Science 300, 2065-2071
25. Ng, K. K.-S., Drickamer, K., and Weis, W. I. (1996) J. Biol. Chem. 271, 663-67426. Feinberg, H., Guo, Y., Mitchell, D. A., Drickamer, K., and Weis, W. I. (2005) J. Biol. Chem. 280,
1327-133527. Sanders, R. W., Venturi, M., Schiffner, L., Kalyanaraman, R., Katinger, H., Lloyd, K. O.,
Kwong, P. D., and Moore, J. P. (2002) J. Virol. 76, 7293-7305
28. Calarese, D. A., Lee, H. K., Huang, C. Y., Best, M. D., Astronomo, R. D., Stanfield, R. L., Katinger, H., Burton, D. R., Wong, C. H., and Wilson, I. A. (2005) Proc Natl Acad Sci U S A102, 13372-13377
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
9
Footnotes and abbreviations*This work was supported by grant GM50569 from the National Institutes of Health to W.I.W and grant075565 from the Wellcome Trust to K.D. Some of this work is based upon research conducted at theStanford Synchrotron Radiation Laboratory (SSRL), a national facility operated by Stanford Universityfor the DOE, Office of Basic Energy Sciences. The SSRL Structural Molecular Biology Program issupported by the Department of Energy, Office of Biological and Environmental Research and by theNational Center for Research Resources, Biomedical Technology Program and National Institute ofGeneral Medical Sciences, NIH. Part of this research was carried out at the Advanced Light Source,Lawrence Berkeley National Laboratory, which is supported by the DOE.
Coordinates and structure factors have been deposited in the Protein Data Bank, ID 2IT5 for theDC-SIGN–Man6b complex and ID 2IT6 for the DC-SIGN–Man2 complex.
4The abbreviations used are: CRD, carbohydrate-recognition domain; HIV, human immunodeficiencyvirus; Man2, Manα1-2Man; Man6b, Manα1-2Manα1-3[Manα1-2Manα1-6]Manα1-6Man; Man3GlcNAc2,GlcNAcβ1-2Manα1-3[GlcNAcβ1-2Manα1-6]Man; Man4, Manα1-3[Manα1-6]Manα1-6Man.
Keywords: DC-SIGN, high-mannose oligosaccharide, C-type lectin, carbohydrate recognition, crystalstructure
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
10
Figure Legends
Figure 1. N-linked high-mannose structures. The full nine-mannose structure (Man9) is shown in thegreen box. The outer trimannose moiety, marked with a black box, is present in both the Man6a (red box)and Man6b, (blue box) fragments.
Figure 2. Electron density maps for bound ligands. The indicated bound ligand orientation is shownsuperimposed on the Fo-Fc electron density map (green, 2σ contour) calculated from a model from whichthe indicated orientation was omitted, but which included the alternative orientation. A, Man6b majororientation. B, Man6b minor orientation. C, Man2 major orientation. D, Man2 minor orientation.
Figure 3. Binding of Man6b to DC-SIGN. The protein is shown in cyan and the carbohydrate in grey,with carbon, nitrogen, oxygen, and calcium represented as white, blue, red, and green spheres,respectively. Hydrogen bonds are shown as dashed gray lines, Ca2+ coordination bonds are dashed blacklines, and key van der Waals interactions are indicated by dashed blue lines. Distance criteria forhydrogen bonds and van der Waals contacts are: 2.5-3.2 Å for hydrogen bonds, 3.8-4.1 Å for aliphaticcarbon-aliphatic carbon contacts, 3.7 Å for aliphatic carbon-oxygen contact, aromatic carbon-oxygen 2.9-3.4 Å, aromatic carbon-aliphatic carbon contact 3.6-3.8 Å. A, Major orientation, showing all three visiblesugar residues and their linkages. The positions of the links to the disordered sugars are indicated witharrows. B, Close-up of mannose bound to the principal Ca2+ in the major orientation. C, Minor orientationshowing both visible sugars. D, close up of mannose bound to the principal Ca2+in the minor orientation.E, Superposition of the major orientation on the Man4 structure (5). Man6b is shown in grey, Man4 inyellow. F, The major and minor orientations, shown in grey and yellow, respectively, were superimposedby aligning the pyranose rings at the principal Ca2+ site.
Figure 4. Binding of Manα1-2Man to DC-SIGN. The protein is shown in cyan and the carbohydrate ingrey, with carbon, nitrogen, oxygen, and calcium represented as white, blue, red, and green spheres,respectively. Hydrogen bonds are shown as dashed gray lines, Ca2+ coordination bonds are dashed blacklines, and key van der Waals interactions are indicated by dashed blue lines. Criteria for assigninghydrogen bonds and van der Waals contacts are given in the Figure 3 legend. A, Major orientation, sideview. B, Close-up of mannose bound to the principal Ca2+ in the major orientation. C, Minor orientation,showing the single visible sugar. D , close up of mannose bound to the principal Ca2+in the minororientation.
Figure 5. Models of Man9GlcNAc2 binding to DC-SIGN. Comparison of an NMR-derived structure ofMan9GlcNAc2 (23) with the major and minor Man6b orientations. The protein is shown in cyan,Man9GlcNAc2 in green, and Man6b, Man4, or Manα1-2Man in yellow. Carbon, nitrogen, oxygen, andcalcium are shown in white, blue, red, and green, respectively. A, Superposition of Man9GlcNAc2 on themajor orientation of Man6b. For clarity the GlcNAc2 moiety of Man9GlcNAc2 is not shown. B ,Superposition of Man9GlcNAc2 on Man4 (5), which corresponds to the major Man6b orientation butincludes the α1-6-linked mannose (see text and Fig. 3e). The loop in DC-SIGN-CRD (residues 367-374)that is in the vicinity of the two GlcNAc residues is shown in orange. C, D, E, Superposition ofMan9GlcNAc2 onto the major Manα1-2Man orientation (yellow), corresponding to the minor Man6b
orientation. C, α1-3 branch terminus of the outer trimannose core. D, α1-6 branch terminus of the outertrimannose core. E, terminus of the α 1-3 branch of the inner trimannose core.
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
11
Table 1. Crystallographic statistics for DC-SIGN CRD ligand complexes.
Man2 Man6b
Data collection
Space group P43 P43
Unit cell parameters a,c (Å) 55.64, 53.20 55.96, 53.26
Resolution range (Å) (last shell) 100 - 1.95 (2.06-1.95) 100 - 2.4 (2.46-2.40)
Rsyma (last shell) 6.6 (20.8) 8.5 (19.8)
% complete (last shell) 99.9 (100) 99.6 (99.8)
Average multiplicity 4.7 4.6
Mean I/σ(I) 18.1 (7.2) 15.1 (7.6)
Refinement
No. reflections working set 11379 6198
No. reflections test set 572 303
No. protein atoms 1071 1071
No. ligand and solvent atoms 170 116
Rfreeb 0.241 0.252
Rb 0.196 0.198
Average B (Å2) 25.8 27.3
Bond length rmsd (Å) 0.005 0.006
Angle rmsd (°) 1.22 1.22
Ramachandran plot: (% in most favored/
allowed/ generous/ disallowed regions)
88.8/ 11.2/ 0/ 0 89.6/ 9.5/ 0.9/ 0
aRsym = ∑h∑i (| Ii(h) | - | <I(h)> |) / ∑h∑i Ii(h)where Ii(h) = observed intensity, and <I(h)> = mean intensityobtained from multiple measurements. bR and Rfree = ∑ ||Fo|-|Fc|| / ∑|Fo|, where |Fo| = observed structurefactor amplitude and |Fc| = calculated structure factor amplitude for the working and test sets,respectively.
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Feinberg et al.
12
Table 2. Competition data for ligand binding to DC-SIGN and DC-SIGNR CRDs.
DCSIGN DC-SIGNR
Ligand KI KI vs mannose KI KI vs. mannoseµM fold increase µM fold increase
Man 2300 ± 100 1 2500 ± 200 1Manα1-2Mana –a 4.1 ± 0.1 –a 3.1 ± 0.5Man6a 183 ± 18 12 ± 3 277 ± 22 10 ± 2Man6b 157 ± 17 14 ± 1 251 ± 28 11 ± 2Man9 73 ± 6 32 ± 4 128 ± 17 20 ± 5Man9GlcNAc2 26 88 54 43
aRelative values of KI vs. mannose for Manα1-2Man taken from Ref. 4. Note that the absolute values ofKI in those experiments cannot be compared to the values shown in the rest of the Table as the solid-phase competition assays were done with a different batch of iodinated Man-BSA.
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from
Hadar Feinberg, Riccardo Castelli, Kurt Drickamer, Peter H. Seeberger and William I. Weisglycans found on viral glycoproteins
Multiple modes of binding enhance the affinity of dc-sign for high-mannose n-linked
published online December 6, 2006J. Biol. Chem.
10.1074/jbc.M609689200Access the most updated version of this article at doi:
Alerts:
When a correction for this article is posted•
When this article is cited•
to choose from all of JBC's e-mail alertsClick here
by guest on February 16, 2018http://w
ww
.jbc.org/D
ownloaded from